Random array of scripts to price securities, analyse market data, etc..
file: stock_correlation.py
Find correlations between pairs and graph the highest.
Use SEC 13F form data to create a portfolio of securities traded by hedge funds/institutional traders and then benchmark its returns against the S&P 500. Securities held more often by multiple funds are weighted higher.
Data from Fintel
file: short_interest_tsla.pySP800-22 Rev 1a PRNG test suite
Use time series analysis to predict optimal exit and entry of short positions using FINRA/NASDAQ TRF Short Interest data from Quandl.
file: yield_model.py
Regreeses Federal Reserve model parameters to predict futures values. Data can be found here. anomalies
file: sentiment_news.py
helper files: process_sentiment.py, get_news_sites.py
Uses newspaper to grab headlines and uses the flair NLP library for sentiment analysis to find the overall positivity or negativity of the news. Will attempt to integrate it into trading strategies and asset pricing models.
Looks at the Kaggle stock dataset to find patterns in the aggregate levels of anomalies over time. prophet is used to try to find a time series trend.
Graph of the log(daily percent change) of a single security with in-sample outliers in red and out-of-sample outliers in green.
Total of all outliers over time.
prophet model on the anomaly index.
file: random_test.py
Uses PRNG tests on log returns to try to test the randomness of market returns. Requires this great repository for the NIST tests.
Sample output:
/home/andrew/Documents/dev/econ/market_random/sp800_22_tests/sp800_22_random_excursion_test.py
J=3
x = -4 chisq = 0.428871 p = 0.994500
x = -3 chisq = 0.600144 p = 0.987997
x = -2 chisq = 1.000300 p = 0.962542
x = -1 chisq = 0.999700 p = 0.962590
x = 1 chisq = 4.333033 p = 0.502529
x = 2 chisq = 8.247775 p = 0.143099
x = 3 chisq = 0.600144 p = 0.987997
x = 4 chisq = 0.428871 p = 0.994500
J too small (J < 500) for result to be reliable
success = True
plist = [0.9945000204954214, 0.9879969685847894, 0.9625415766568545, 0.9625899708254657, 0.5025287424472648, 0.14309859788329615, 0.9879969685847894, 0.9945000204954214]
/home/andrew/Documents/dev/econ/market_random/sp800_22_tests/sp800_22_serial_test.py
psi_sq_m = 2.8
psi_sq_mm1 = 1.2
psi_sq_mm2 = 0.4
delta1 = 1.6
delta2 = 0.8
P1 = 0.808792135411
P2 = 0.670320046036
success = True
plist = [0.8087921354109985, 0.6703200460356384]
/home/andrew/Documents/dev/econ/market_random/sp800_22_tests/sp800_22_maurers_universal_test.py
sum = 7.16992500144
fn = 1.19498750024
success = True
p = 0.0314262987784
etc...
main file: sp500_tech.py
Creates a VAR model of 11 different tech stocks and predicts future price movement.